Principal Security Researcher - Red Team

Microsoft • Redmond, WA

1d • $142,800 - $304,200

About The Position

Microsoft Security is hiring a Principal Security Researcher for a new red team focused on agent security, offensive security, cloud identity, attack automation, and adversarial AI. This individual-contributor role involves hands-on red teaming of agentic systems and AI-enabled security products. The researcher will design realistic attack chains across models, tools, identities, permissions, and isolation boundaries, then turn critical findings into durable improvements such as regression evaluations, automated attack suites, stronger controls, and safer product designs.

The role calls for depth in offensive security or agent security, fluency across cloud and AI systems, and the judgment to produce reproducible evidence that engineering, product, security, and independent review teams can act on. Redmond, Washington is the preferred work location; remote work within the United States may be considered for highly qualified candidates.

Microsoft describes its mission as empowering every person and organization to achieve more. Its stated values include growth mindset, innovation, collaboration, respect, integrity, accountability, and inclusion.

Requirements

• Doctorate in a relevant field and at least 3 years of related research experience; or master's degree in a relevant field and at least 4 years of related research experience; or bachelor's degree in a relevant field and at least 6 years of related research experience; or equivalent experience.
• Must be able to meet Microsoft, customer, and/or government security screening requirements, including passing the Microsoft Cloud Background Check upon hire or transfer and every two years thereafter.

Nice To Haves

• Hands-on expertise in offensive security, agent security, cloud or identity security, application security, exploitation, or a closely related discipline.
• Experience designing security investigations that produce reproducible technical evidence and actionable remediation.
• Programming and engineering ability sufficient to build attack tooling, test harnesses, exploit prototypes, and automated security tests.
• Experience testing large language models, agentic systems, tool use, plugins, connectors, multi-agent orchestration, or AI-enabled security products.
• Experience with indirect prompt injection, cloud identity, authorization, sandboxing, container security, web or application exploitation, or offensive-security operations.
• Knowledge of secure agent architecture, least-privilege design, isolation boundaries, human-in-the-loop controls, and high-fidelity security evaluation.
• Demonstrated ability to explain complex attack chains, severity, and residual risk to engineers, product teams, and senior decision-makers.
• Experience converting security findings into production mitigations, automated regression tests, architectural changes, or durable security controls.

Responsibilities

• Design and execute end-to-end attacks against agentic systems, AI-enabled security products, orchestration frameworks, tools, plugins, connectors, identities, permissions, and isolation boundaries.
• Investigate indirect prompt injection, unsafe tool use, scope violations, privilege escalation, confused-deputy behavior, secret exposure, sandbox escape, cross-agent manipulation, and tampering with human oversight.
• Build realistic attack chains using attacker-controlled telemetry, alerts, code, web content, and other inputs encountered by agents in production workflows.
• Develop automated attack suites and high-fidelity test environments that continuously exercise model behavior, harness logic, tools, identity, and representative product telemetry together.
• Create rigorous threat models, attack hypotheses, and reproducible evidence; assess exploitability, severity, and customer or governance impact.
• Partner with engineers, model researchers, and product security teams to identify root causes and drive findings to verified remediation.
• Convert confirmed weaknesses into regression evaluations, adversarial scenarios, automated tests, stronger authorization boundaries, and safer system designs.
• Communicate attack paths, findings, and residual risk clearly to technical teams, leadership, and independent reviewers.
• Track emerging agent attack techniques and real-world adversary behavior, then evolve MART methods, tooling, and investments accordingly.

What This Job Offers

• Job Type: Full-time
• Career Level: Principal